Signature Block Truck Driver Application Made Easy

Signature block truck driver application

you can't do the interesting things that ethereum wants to do without allowing open-ended computation you don't want to do the interesting things that ethereum wants to do in bitcoin because you don't want to allow the additional computational complexity hey folks on today's episode of speaking of bitcoin we'll dig into a critical topic in the blockchain world the fees we pay to get our transactions recorded and the distributed permanent record and some interesting differences between the top two cryptocurrencies bitcoin and ethereum today as always i'm joined by the other host of the show stephanie murphy hi jonathan mohan hey hey and andreas m antonopoulos hello thanks to all the hosts and to you the listeners for sitting in on today's session so transaction fees it's not a sexy topic but it's a critically important one under normal circumstances it's like putting a stamp on a letter it's generally quite cheap and how long it'll take for that 15 bucks worth of bitcoin or that cryptokitties token to arrive at its destination is fairly predictable and to be clear normal means that the number of transactions being sent is less than the total capacity of whatever blockchain you're using in that situation miners are incentivized to take whatever fees they can get and fill up each block with as many transactions as they can sometimes even transactions that have incredibly low or even no fees at all get swept along but all that changes when usage spikes up and miners have to start picking and choosing which transactions will get confirmed quickly and which ones they're willing to wait on when that happens senders start to compete for faster service by increasing the amount they're willing to pay which can spiral upwards in expense as the market struggles to figure out exactly what it's worth to them for that fast service so we've got a few areas to touch on in today's discussion but let's start with basics andreas what's the thinking behind this two-phase approach to fees and what are we trying to accomplish here the simple answer here is that there will always be capacity constraints in a blockchain especially a blockchain that has decentralization as its goal and therefore needs to keep the resource utilization low enough that many people can participate in either the validation and mining or just the mining of transactions so in order to do that you have constraints on how much data can be recorded in the forever immutable blockchain because it has carrying costs then the question is how do you allocate that resource and the simple answer to that is that there's two ways of allocating that scarce resource which is the capacity of the blockchain which is not infinite one way is a market an auction effectively which is what fees do and the other way is by authorizing someone to decide what should take priority and what shouldn't and that second part is often hand waved meaning that people assume that it's okay we can just have free transactions or very very cheap transactions for everything but if you really do have a resource constraint which you do if you're trying to keep things decentralized eventually you're going to have applications being developed that try to use that for different purposes and then if someone has to decide the power to decide which applications are worth transmitting and which applications are not worth transmitting is a pretty important power it effectively leads in my opinion to a dictatorship of the developers they get to say this is spam this is real this is useful this is not a market-based approach says that the best mechanism for allocating scarce resources is a market and by having a fee auction the sender of a transaction can tell the market how much they value the prioritization of their transaction from their own perspective and make that choice if it's not important to them they can put a low fee if it's very important they can put a high fee but nobody gets to tell them that their application is or isn't important you know this reminds me of i believe it was like the original invention of proof of work where it was a solution to email spam because there was some problem where people were sending out email spam and you know was it adam back who was inventing this that's right cash cash exactly okay so hash cash was meant to basically prove that you are a human being whose computer is in work in order to send out an email in order to prove that it's not just a spam transaction and there's a cost to sending an email or at least if you want to prove that it's a legit email there's a cost to it instead of you know at the same time the congress was trying to make laws about what is spam and you have to have a physical address in order to spend out email newsletters and stuff like that and they had the canned spam act but hash cash was a market solution to the very same problem yes and ultimately that is really the choice in fact in most of this debate the idea of this being magically resolved through decisions being made elsewhere is the problem because if you don't have a market all other choices that you make create some kind of centralization of power you can't get around the fact that there is a scarce resource and if you increase or if you distribute that resource too liberally if you create a very big supply of that resource you're just shifting the problem to a decentralization problem if you make gigablocks you can achieve very very low fees almost zero or perhaps even zero but the problem is then you're just shifting the carrying cost to miners and validating nodes and making it so that it is so expensive to validate a transaction and a blockchain that fewer and fewer people can do it which centralizes power in the hands of those people so if you replace the market with alternative solutions basically you try to do some kind of centralized directed management of this problem you don't replace the problem the problem doesn't go away you just shift it into some other form of centralization i want to draw us back from kind of the blockchain portion of this conversation for a second because i think that there actually really is a real world analog that demonstrates that this is a critical problem that even when you have those sort of authoritative types of solutions doesn't work very well i think we're all probably at this point and many of our listeners common with this idea of robocalling right where in some cases certainly for me the vast majority of the phone calls that i get to my various numbers are not real people they are essentially you know companies that are illegally marketing to me and it is against the law and has been against the law for some time but from a system standpoint there's no real mechanism to stop this so it doesn't matter that it's against the law because the enforcement is kind of like it requires a lot of work on the person who is the victim of it in order to actually make anything happen and the cost to actually do it i put together a proposal a year or two ago about adding a cost to this and i figured out that to robocall all of seattle twice it would be something like 200 wow and so again it's an example of like there is an economic cost there but it's so low that it doesn't accomplish the purpose of keeping the network clean and in fact you know there really is no meaningful solution without completely reworking the way that we do phone calls in order to solve that problem so i mean that's not a blockchain example but i think it's another example of why using market-based solutions or some kind of solution that doesn't rely on it just being against the law is very important even in centralized systems i have a solution please it's called do not disturb i turned it on in 2013 it's still on i know right but that's my point is like i still pay for three phone lines but i don't pick up the phone anymore yeah that's the thing it's been completely ruined i mean the do not disturb that i have is turning my ringer off right exactly yeah exactly that's the solution you turn your ringer off you no longer allow people to asynchronously interrupt your time oh i thought you were talking about the government do not call list no no i'm talking about turning my ringer off in 2013 and leaving it off yeah right same here i mean that's because everyone knows that this do not call list is useless and it doesn't work which is why the world now uses zoom calls instead of phone calls because who picks up the phone anymore but also what kind of freak would call you rather than signal voice message you right that's my new human check in crypto is if you want to reach me just call me through signal i immediately know the difference between a bot and a human and it's because oh wow use the most basic form of encryption and all of a sudden you get back to meat space related interactions again yeah what this really demonstrates is that you can't replace economic incentives with regulation and you most certainly can't negate real economic incentives and costs with even more regulation and in the case of a fee market that's exactly the issue which is that you have to regulate the allocation of the scarce resources and you can either do it with economic incentives and marketplaces or you can pretend that the economic incentives don't exist and try to regulate a solution that is kind of compatible and it will always be inferior and transfer power to those who have the regulatory capacity or you can go even further and try to negate the economic reality with regulation and then you end up with a system that gives even more power to regulators and ultimately doesn't solve the underlying problem the problem being that it costs 200 to spam all of seattle or that you know you transfer power to miners and reduce the number of validation nodes if you allow for very very large blocks but then when you get to regulating your system it comes down to what are you prefacing and what are you trying to dissuade and what are you incentivizing what are you dissentivizing and it's not always so clear cut or even comes down to like a moral category of well you know gigabit blocks would preclude the little guy it really then comes down to philosophy about your network and what specifically you're trying to engage with and promote yes and you know i think that's a valuable discussion to have it's a valuable discussion to have about what are the principles and what are you aiming to achieve and therefore which side of these design tradeoffs do you want to fall on how do you want these optimizations to trade off various principles the problem i have is that very often this is couched in a way that there are no design trade-offs so you can have your cake and eat it too that there is such a thing as a free launch that either resources are unlimited or economic incentives don't actually apply well can't we just pull in ethereum and say that gas fees are really bad right now and call that gas fees 1.0 and then give a really good name to gas feast 2.0 like let's say i don't know something that rhymes with casper and then call that gas fees 2.0 and then put all of the non-feature complete solutions that have no trade-offs in that asphy 2.0 solution as we ultimately never arrive to receiving gas feed 2.0 oh then you'd just be full of gas i mean or full of ether okay so talking about fees i mean like again this is not a problem you know jonathan's specifically bringing up that ethereum is having a problem with fees at the moment or was until recently i think drawing back from that this is a problem that's affected any blockchain that's actually gotten sufficiently popular enough to kind of overcome that capacity limit right in terms of the once you have more transactions coming in or more activity on a blockchain than a blockchain can support during that same amount of time then you have a problem that basically escalates until people are kind of forced to choose not to transact because the cost of doing a transaction the cost of interacting with the blockchain is too expensive for them i think that there's a valid conversation to be had there about whether or not that's a desirable feature of blockchains but i think to andreas point i don't actually think there's another way to do it like all of the other alternatives seem like they're kind of worse in a lot of ways it depends on what you're trying to achieve i think the thing that we're missing yet from this conversation is that when what you're talking about is akin to the post office you know send a letter right and do you want to pay ups express or are you willing to just go first class or do you want you know donkey mail the cheapest possible eventual possible delivery maybe not that's a fine conversation to have the side effect of fee markets and this is where things get tricky is when you have things that absolutely must be delivered because there is a deadline let's use a ridiculously out of this world example let's say you have an election and someone's trying to mess the post office up so that people can't actually deliver the mail on a very specific deadline which is when you need to count the votes now in this ridiculous example you can't solve that with a fee market because then you've created a situation whereby something that should have guaranteed execution now is dependent on whether you can afford the fee at the time and sometimes you can't change the fee now this applies in very obvious cases in both ethereum and bitcoin when you start talking about second layer protocols or smart contract execution and we've seen this happen in two areas it becomes a problem for fee markets one is lightning network let's say you have a commitment transaction that allows you to effectively broadcast a penalty if the other party tries to cheat well you have a specific period of time in which you need to broadcast that penalty there's a timeout mechanism right if you don't broadcast your penalty transaction within that time frame the other party gets to steal your channel balance or part of your channel balance and so the game theory the mechanism of lightning depends on your ability to send that penalty transaction but the way it is today you have to calculate the fees in advance what if when the time comes and you need to send that penalty transaction fees on the network are out of whack that's problem number one you can't recalculate the fee because that requires changing the transaction the second problem of course is that even if you can recalculate the fees and increase the fee of that penalty transaction through some mechanism enlightening that would be a thing called anchor payments which is essentially child pays for parent then you now have a very difficult game theory situation what if the fees that you have to pay to get this penalty transaction in are more than the penalty that you're going to earn more than the reward you get from penalizing the other party well then it's not worth doing so you end up losing after all and of course you might not be able to reach any amount of fee that gets that transaction in the backlog may be so huge that no matter what you try to do you can't really do it so that's one situation the other situation similar in ethereum was again deadline based so during the big crash in march where the price of all cryptocurrencies well almost all dropped by anywhere from 40 to 60 percent a bunch of die collateralized loans became under collateralized now that means those loans go to auction if they're not re-collateralized as quickly as possible so you have a very short window of time less than a day to recollateralize those loans before they go into a secondary market auction and get auctioned off and in this particular case the gas fees exploded at the same time that all of these loans became under collateralized and even though people were ready and willing and they had these available to recollateralize these loans they couldn't get their transactions through and so their loans went to the secondary market to get liquidated about five and a half million dollars lost and i don't know if they ended up getting refunded they did not they did not right they took a vote not to that's right yeah the original thought was and i believe there was an initial vote that suggested that they would actually be compensated for those losses but my understanding is that in the final vote it did not wind up that way and that just wound up being losses that were eaten by the people who were affected by it right so that's another example whereby there are some problems with the free market approach or some side effects of a free market approach where if you have a few markets and you don't have a mechanism to prioritize certain must-do transactions then that fee market can result in failures is that really a problem with the fee market approach or is that simply that there was again even in a fee market approach so much competition to have one of those transactions that actually does happen during this incredibly busy time when so many people are seeking to use the network yes is that a failure of the fee market or is that just a reality of the fee market that fee markets necessarily have winners and losers well i would even go one step further and say that fee market or not blockchains aren't order books i think we keep running up on this constraint of let's entirely adapt the fundamentals of finance that are predicated on an order book on a system that batches transactions not utilizing atomic ordering and so when sh breaks it only breaks when everything is breaking and so everything seems optimistic until one black swan occurs and it's only when a black swan occurs that everything catastrophically fails because it's not an order book it's a batching system and you know it's sort of like the early days of amazon where people were just pushing their server to amazon and then saying this is not cheaper i have no idea what's going on they had to fundamentally re-architect their design to optimally use amazon cloud and then now people understand inherently how to architect to that if you're building a system because you're lazy and most of the thinking in ethereum dapps is fundamentally lazy you can't build a system that requires an order book when sh goes upside down because that's not ethereum and that's not bitcoin right and so you know that's one of the ways to test these systems is every now and then you do have an event like that and everybody gets to rethink their original design and of course also their original risk models and you know how much excess collateralization do you need in a die loan with the new assumption the new assumption being that you can have a 55 price meltdown and gas cost explosion on the same day which of course are going to happen on the same day so when it rains it pours in blockchains and that's when these models are strongly tested but there's some very interesting work happening as a result of this including to how fee markets are structured so far we've talked about an auction model it's a weird type of auction though because it's a blind auction and you're operating in retrospect so you can only see the bids that you can see at the moment but there are other secret bids that you haven't seen yet and the auction runs continuously until some arbitrary moment in time when suddenly all of the bids are executed so it's a very weird kind of auction and a lot of wallets have a lot of difficulty dealing with this you have no way of knowing what the minimum fee will be to get into this block or the next several blocks you can't predict and it's a feedback loop whereby if you pay more in order to get in you're also changing the market causing others to pay more in order to get in and that ends up being a race to the top i guess this also comes down to a broader topic in relation to system design when it comes to variants you know it's sort of like a variance is a game of hot potato and you have to decide in your system at all times who's the one holding it and for how long and it seems that even more so in ethereum than bitcoin that being able to predictably understand how much executing a transaction should cost has enormous variance from time to time and you know as the system scales and as it potentially has more users than people who bake bread um is how much variance how much of that system variance should we maintain that the individual user should continue having as they interact with anything in ethereum when we're talking about you know in the postal service i could be mailing a letter god forbid i would ever do so from the northern part of manhattan to the southern part of manhattan and let's say stamps cost 40 cents that would cost me 40 cents but america's decided that the same person mailing the same letter from alaska also costs 40 cents so i think that you know when we look at transaction fees especially in ethereum there's some real conversations to be had about you know user variants both in terms of time and order but also just in terms of cost and predictability you know we talk about mining pools aggregating in order to lower like these tiny tiny amounts of variance and yet we're very far behind in lowering variance when it comes to the user's experience in transaction fees so we've talked about how to handle fees and just these problems of allocating resources on a network like a cryptocurrency network broadly but let's get more specific and talk about the two best maybe known or most used cryptocurrencies bitcoin and ethereum which have very different ways of handling this problem of fees and resource allocation so with bitcoin it's pretty straightforward you're sending a transaction and you attach a fee to it in bitcoin ethereum has this thing called gas now like what is gas how does it work and they're even talking about changing it recently but let's talk about how it works right now first so the fundamental difference between ethereum and bitcoin in terms of the fees starts much earlier and has to do with the purpose of fees in the two systems and it goes back to the fact that bitcoin is not touring complete and ethereum is turing complete turncomplete means that it can do arbitrary computation and that arbitrary computation can go into infinite loops and last forever that's a very vast simplification of course and you can't really predict how long something is going to take to execute without basically executing it at which point it might end up taking forever to keep it simple with bitcoin but the scripting language is not turn complete so you can actually calculate exactly how long it's going to take to execute and it has a maximum limit on how many signature operations or sig ops it can execute in a script that has rarely reached it's a very very big limit and it's irrelevant for the vast majority of purposes which means that transaction fees in bitcoin are really for the effort of mining not the effort of validation most transactions validate in more or less the same amount of time and it's not a huge amount of time with ethereum every time you validate a transaction whether you're a miner or not you have to validate its entire execution path you have to run the transaction any smart contracts it calls the execution of those smart contracts the changes they make to the data store retrieving and saving data et cetera et cetera et cetera and these contracts may call other contracts which may call other contracts which may loop you never know how long that's going to take the purpose of gas in ethereum is to make sure that you have an upper limit so it's there not just to prioritize the inclusion of transactions in a block it's primarily there to constrain the amount of computation that a validating node has to do by setting a ceiling on it and that's the maximum amount of gas that you can put in a transaction and the maximum amount of gas that you're willing to spend which is what you specify when you run a transaction so if i send an ethereum transaction i am specifying within that ethereum transaction what my maximum amount of gas is and the maximum price i'm willing to pay for that gas and the price is calculated in gigaway which is ether so gas isn't a currency itself it's like a pseudo currency that is denominated in ether with a variable exchange rate and it's used not to prioritize transactions in the blockchain but primarily to constrain the amount of compensation that has to be done by everyone to validate those transactions by running all of the smart contracts okay so let me just make sure i understand what you said there andreas so in bitcoin there are a limited number of defined things that you can do and you can take those defined things you can create combinations of them that may involve some complexity but ultimately because there are only a couple of things that you can do the complexity is limited and sort of that also means the types of things you can do are limited you can't have fully functional smart contracts in the same way that you can in a turing complete system like ethereum but it also makes it kind of much more predictable and much more standardized in terms of what type of activity a miner might actually see and wind up processing and every bitcoin script has a fixed runtime it doesn't depend on anything else it will take as much as it needs to take to evaluate that script depending on how many script commands there are and how many signature operations you do in the script but that's it the script can't loop back on itself and start running again and it won't change how long it runs depending on something that's outside of that script whereas in ethereum it will it will run a smart contract that can loop and the length of that loop may depend on a factor that's completely outside of that transaction that has to do with another transaction that somebody else recently did so when we think about this is it correct in thinking that bitcoin is a more rigid but predictable system whereas ethereum is a more flexible but sort of unpredictable inherently system exactly and predictability is the basis of the turing problem which is whether you can predict the execution length or computation required for a specific program the during theorem says that you can't if the language is during complete then you can't model with another computer how long it's going to take to execute so and this really speaks to kind of the purposes of what these blockchains are for where sort of bitcoin really at its core is attempting to be a kind of money right that may have systems built on top of it that had lots of complexity ethereum really does want to support that complexity at the base layer so that you don't have to build additional systems on top which come with their own sets of complexities you can do the interesting things that ethereum wants to do without allowing open-ended computation you don't want to do the interesting things that ethereum wants to do in bitcoin because you don't want to allow the additional computational complexity and that's the fundamental design trade-off there okay so if you understand the gas is basically how you meter the amount of computation then the next step is to understand that every command that you ask the ethereum machine to run has a specific cost in gas as the accounting system and so if you want to add two numbers together that costs two gas let's say i'm just picking numbers randomly here if you want to do a hash that costs 200 gas if you want to store data that cost 250 gas if you want to do an elliptic curve multiplication that costs 2 000 gas and you know proportionally these are set in advance to make some kind of sense in terms of how that reflects the amount of computation that's required and they've been tweaked from time to time both because some things were underestimated and gave opportunities for denial of service and other things were overestimated and they wanted to make those more appealing so there's essentially a price list for every processor command you want to run there's a price list that tells you how much that costs and when your smart contract starts running when your transaction runs a smart contract the ethereum virtual machine wherever the transaction is being validated basically starts keeping a tab of how much your gas you're spending keeping an eye on the total amount to see if it exceeds the maximum you've agreed to pay in the transaction if the system completes successfully your smart contract runs you get charged for that gas interestingly enough if your transaction fails you still get charged for some gas because your transaction is still mined and it produces an error condition or something like that you still pay some gas although perhaps not the maximum amount of gas for that transaction and one interesting wrinkle there which opens up quite a few possibilities is that there is one operation that actually has a negative price and that is the operation of releasing data allocation so when you record a variable or you store data on the ethereum blockchain you have to pay for that but when you release that data you actually get a small amount of gas back meaning the gas is released sorry the cast is released had to get a fart joke in there somewhere yes exactly i mean it was begging for it was it was so not to make us think about this but the bottom line i couldn't help it i couldn't help it so the bottom line is that because this one operation gives you a credit which is not going to be enough to make your transaction free but can reduce the overall cost of a transaction if you're releasing enough data in that transaction you have some interesting side effects and that is you can actually have a market for gas that emerges from this process okay so how do you get a market from gas is that because the price of gas is denominated in ether which fluctuates based on the market prices and you can sort of bank it when it's cheap and then use it later when it's more expensive exactly so take a smart contract for example that has two functions in it and one is let's say bank gas and the other one is withdraw gas right and what the bank gas does is it allocates an array of a thousand bytes so it just goes and allocates and allocates and allocates and allocates a thousand bytes and that costs you two hundred thousand gas because that's 200 times each byte that you allocate i'm picking numbers out of thin air right now just to make the point now if you do that if you run that part of the function when gas is really sheep when the exchange rate of gas to ether is really low you end up allocating 200 000 units of gas at a very low price now when gas is expensive you can go and run the other function and you can run it as part of another contract execution you can just prefix your contract execution wrap effectively the smart contract that you want to do which is something useful a decentralized exchange a d5 project whatever you can wrap it with a call that first releases the banked gas that you had withdraws it and what that function does is it goes and it releases that memory allocation giving you a partial gas credit which then goes towards your overall gas costs for running the other thing you were doing so let's say you were using it to withdraw some die from a smart contract and normally that would cost you 200 000 gas but instead because you first withdraw gas from this bank to gas contract it only costs you 50 000 gas to run this thing because you pay 200 000 to withdraw your die but you get 150 000 credit from releasing the banked gas and so in practice it only costs you 50 000 gas so you've bought gas when it's cheap and you've effectively gotten it back when it's more expensive and that way you have essentially a gas savings market okay so like do people actually do this do they create like crap contracts that tie up a lot of gas and then release it later absolutely well they don't create crap contracts to do that because you only need one so once someone created a contract that was nice enough simple enough easy to audit etc that did this process and had the ability to bank gas when you need it when it's cheap and then release that as part of a bigger execution of contracts when you want to get a credit uh everybody started using that you know there's no reason to do another contract you can just use the contract that's already there and it's got a nice api and it's very easy to write smart contracts that include this in their execution and this is used in a number of different smart contracts to save on gas right and this comes back to like system philosophy which is whether or not you think this is a good thing or a bad thing because you know the reduction in variance on gas prices well that means that by definition lowering the upper variance and the lower variance which one way to look at it would be great we're creating certainty in ethereum gas at least increasing it and on the other side of saying those evil capitalists they're increasing the base fee of ethereum writ large and we need to do something about it let me make sure i understand exactly what's happening here in terms of like practical real world impact right so if a transaction that i know i'm gonna make in the future you know it costs two dollars today but it might cost twenty dollars you know the day that i want to execute it you have the perception of it costing 20 in the future or the anticipation of it potentially costing 20. right exactly like i know that there are transactions that i'm going to want to make during a time when it's high volume you know the currently blocks are not full and i can get an equivalent amount of data into the blockchain for like two dollars so effectively i do that i pay the two dollars now and then at the later point when i'm getting rid of this data that i've put in but transactions cost twenty dollars i'm actually getting ten dollars worth of value back in terms of that rebate so i've had a gain there effectively of about nine dollars for the one dollar i wound up actually paying to think about it in the most real world examples it's sort of like deliverable futures in the commodities market yeah it's a futures market right but deliverable futures where you run a trucking company and you just need to lock in the price of diesel and it doesn't matter if you're paying a slight premium for it you need to lock it in because you just need to know what your margins are going to be for the next six months for the next two quarters or whatever and that's what you need to do in order to run a commercial enterprise now someone could look at that and say that's fantastic this guy is able to run his company effectively another person could look at it and say that dirty mo increasing the market price of gas and his anticipation of the price going up into the purchasing and locking down of these futures is in and of itself what's increasing the price of gas itself evil speculators are messing with the market yes you get that kind of effect it's a secondary effect of this market your anticipation that the price will go up actually does cause the price to go up because other people can see that bid you're making and are probably trading against those futures in the opposite direction if they don't think it is going to go up or in the same direction if they think it is and so you've effectively created a secondary market which then people start doing naked speculation not because they're trying to hedge the price of diesel for their trucking company but because they're trying to make a buck off other people hedging the price of diesel for their actual trucking company and in some cases are artificially increasing the price of diesel so yes that's exactly what happens interestingly there's a new proposal now to change even more radically the way the gas market works by effectively dynamically adjusting the price of gas by burning it and not giving it to miners which is a proposal called eip 1559 which is getting a lot of attention now i don't know what the status is in terms of its implementation i think it's still a proposal and in draft state but it's quite vigorously pursued right now and there's a lot of interest in it so the point of this proposal would be to stabilize the price of gas so that it really doesn't have these fluctuations that can be harvested later yes it's basically suggesting a change that effectively changes the way gas is used to change the incentive mechanisms in this market so that it's not a first price auction mechanism um where you have to bid the maximum fee and the miners get the difference because that encourages as we've seen on bitcoin too the miners to game the system to maximize fees and that's one of the risks of these first type auction systems is that you have a ringer in the auction who's bidding up the price for an item with no intention of buying it as long as they have a relationship with the seller of that item and so if they end up accidentally buying it well they're buying it from themselves so it doesn't really matter right so this is the problem with auctions in the transaction fee space what that is is miners putting transactions onto the blockchain to drive up the transaction fee knowing that if they have a sufficiently large hash rate they're going to get that money back from themselves and even if they don't as long as all of them are doing it they're getting other people's fees while other people are getting their fees so it's all a wash andreas i'm pretty confident that in iceland that's just called banking i'm very confident that in all other markets that's called a cartel [Laughter] and this is a well-known problem so basically here the stated goals and motivation for eip 1559 so the first one is to fix the mismatch between the volatility of transaction fee levels and the social cost of transactions basically the fact that this volatility creates a bunch of inefficiencies in the system because it forces people to over bid and bank gas and all of that delaying transactions for users so needless delays inefficiencies in the auction mechanism and instabilities of blockchains that have no block reward so basically transitioning to fees only and having a possible future instability where fees do not adequately subsidize the security cost so i'm kind of confused as to the proposal as to why it would get adopted because let's just say that casper doesn't get implemented yet and we're still in the ethereum one world which is what the proposal is meant to address how do the incentives work to get miners to signal to adopt a system that would remove their own transaction fee reward the difficulty bomb i'm assuming ethereum has a built-in mechanism that makes mining impossible or increasingly impossible increasingly too difficult over time forcing miners to accept a hard fork at regular intervals to diffuse that difficulty bomb and then reset it and then diffuse it again and diffuse it again and diffuse it again it's a red line in the sand that every now and then has teeth but more often than not doesn't right and the intention here is to not allow miners to take the system hostile and stagnate the introduction of changes by basically cartelizing the blockchain now when it was first proposed i had a lot of skepticism then 2017 happened with bitcoin now i'm thinking it might not have been such a bad idea right we need a different cartel of people who just own a bunch of ether yes so it's all about all of the cartels do you want a cartel of developers a cartel of miners or cartel of owners and what is the relative balance between the three because these power dynamics exist in every blockchain whether you acknowledge them or not and whether or not they're powerful so i can recall an update and geez i wish i could pull it off the top of my head right now there was an uncle rate upgrade that specifically optimized ethereum transactions and reduced the uncle rate but because it reduced the uncle rate it would statistically reduce uh miners ability to generate income off of their uncles what is an uncle is that like an abandoned block or something in bitcoin mining is first past the post the miner who finds the block first wins all of the reward and other miners who may have also found a solution to the proof of work perhaps a few seconds later or even before but weren't seen by the network get nothing in ethereum that's not how it works the first miner gets a proportion of the reward but if other miners also submit additional proof of work those blocks do not get orphaned instead they become siblings of the parent therefore uncles that also get a portion of the reward redistributed to them and share in the reward yeah so the best way to think about it is when millennials make blockchains they get participation trophies yes another way to think about it is it's like you know one winner bingo versus potentially a bingo round that has multiple winners where you know the earlier you win the better the reward is this is an alternative consensus algorithm so if you think the consensus algorithm in bitcoin is greatest cumulative difficulty chain right so the greatest cumulative difficulty chain is the only chain and it wins the parent that introduces the greatest difficulty ends up winning the chain the consensus algorithm in ethereum is called ghost g-h-o-s-t this is why a lot of the future developments are called casper and ghost stands for greedy heaviest observed sub tree greedy because we include additional blocks heaviest that's the cumulative difficulty observed because it actually happens after the first parent is mined and sub-tree because it's not a chain it's a sub-tree and it's an alternative consensus algorithm it's very interesting to read about how ghost works but it does involve this mechanism of uncles now to extend our previous puns if your uncle is producing too much gas then you want to change the number of uncles in the room and stephanie just in case you're wondering the reason why it's called an uncle rate and not an auntie rate is if it was an auntie rate there'd be way less abandonment in the system well in fact an ethereum is called an omer which is a gender-neutral term for our uncles because ethereum is rainbows and unicorns jonathan you should know this [Laughter] so when it came to this proposal that optimized transactions and reduced the uncle rate now i haven't checked in it for a while i'm sure it eventually got passed but miners were like oh you're proposing something that's going to knowingly reduce the amount of reward that we get in the system even if it makes the system more efficient that doesn't really sound like something in our best interest to activate and so we look at you know blockchains being systems of economic incentive it's really interesting to see how you can implement an update that services one context of the ecosystem to the detriment of the party that ultimately approves or votes for the things to get implemented which are the miners yeah i mean it's a battle right it's a battle of interests i mean when i look at this proposal it sounds intelligent i look at it it makes sense i can't help think about you know i'm like how's that conversation actually go because i've worked with miners i've conversed with them i know how they think it's sort of like that scene from the dark knight where the joker just gets them in a room with like a couple hundred million dollars and sets it on fire and it's like welcome to the new world just deal with it how exactly does that conversation go down well pretty much like that because ultimately the money in the middle of the room is the miners invested capital and gpus and vitalik has always stood by with a lighter we're going to burn all of that invested capital and switch to proof of stake anyway so that's that it's a very specific and very deliberate kind of standoff where vitalik from the very beginning did not want miners and ethereum to have the same amount of power that they do in bitcoin and chose to make some very different choices since we've done all of these parent-child uncle relationships i just want to point out that polka dots consensus algorithm which was built after theorems obviously by gavwood is called grandpa and grandpa stands for ghost-based recursive ancestor deriving prefix agreements wow nerds everywhere it's everywhere more acronyms that are actually backrooms yeah i think it's fine but they actually mean something interesting and do tell you a few things about the properties of the system all right so let me go back to eip 1559 because the basic idea here is that there is a base fee per gas which can move up and down each block according to a formula which is a function of how much gas was used in the parent block and the gas limit basically it's like applying a difficulty retargeting algorithm but to gas instead of difficulty so in this proposal eip 1559 gas goes up and down dynamically but based on an algorithm that accounts for supply and demand and in order to prevent the miners from gaming the system the base gas is not given to miners but instead is burned and there is an additional amount of gas on top of the base fee that is a tip to miners which is not considered in the prior decision of a transaction so this essentially disconnects the incentives that miners have to game the system and it engineers the fee market so that the fees cannot vary as much between blocks they have a maximum adjustment between blocks and that's the temp down volatility but effectively what it's done is it's introduced an algorithmic solution in place of an auction based market what are the side effects of that how does the game theory play out well we'll find out when real money is on the table with real actors with interests and incentives competing for that money so you can't really test these systems on paper until you put the money on the table and everybody tries to grab for it yeah that's the thing about ethereum is like everything is a giant experiment and that's exciting but it's also not very stable sometimes i mean that's the thing about this whole industry right it's not just ethereum yes these are real world game theoretic systems that require adversarial execution you need to run them with real money with real adversaries and see what happens and nobody knows in advance exactly what's happened look at the early writings of satoshi nakamoto there was no certainty that this was going to work by any of the people involved not least of them satoshi nakamoto everybody said well we'll see this might work let's see how it plays out and i think the primary philosophical difference between ethereum and bitcoin is not whether it's an experiment or not i think on both sides there's a recognition that it's an experiment the difference is the amount of conservatism in pursuing that experiment and how many things can be broken and sometimes broken on purpose in order to move to the next state of the experiment and in ethereum the culture is very much more progressive in terms of breaking things and in bitcoin it very much isn't and that's great for system of sound money not for a system of smart contracts and vice versa yeah definitely that's a fair point i mean it's all experimental but i don't know there's just something about ethereum to me that just exemplifies this because we see it happen often where things are being tested in the field in the wild and sometimes they break and it's interesting to watch so does this proposed ethereum improvement rely on a smart contract too or does it rely on a hard fork or what well well specifically they're burning the change they're not actually giving it to anyone so i think this is a hard fork change that involves a smart contract but is not implemented solely by a smart contract it's a fairly big change to how the mining mechanism works so i don't think it can be done solely by introducing a smart contract i think we'll all be interested to see kind of how this works out it seems like another attempt at the brave new world of these kind of fee markets and trying to figure out how to make these systems really work in real life in ways that are both kind of fair and actually sustainable over the medium term because they seem oftentimes to be ideas that are at odds with each other yeah in the end you know it's important to keep in mind that this isn't about making something that doesn't break it's about making something that breaks in different ways [Music] one of these days we're going to do a short episode but that's all the time that we have for this episode of speaking of bitcoin thank you very much for listening today's show featured stephanie murphy jonathan mohan andreas am antonopoulos and myself adam b levine this episode featured music by jared rubins and was edited by jonas have any questions or comments send me an email at adam ltbshow.com and if you enjoy the show tell your friends or leave us a review and we'll see you next time